Telecom-Churn-Prediction

A classification project for forecasting the customers who are likely to churn. Random forests, XGBoost, SVM used.

Sections in the notebook:

1. Data Scrubbing/Cleaning: Columns Renaming & Missing Values Treatment

2. Exploratory Data Analysis: Correlation, Univariate & Bivariate Analysis using Visualizations

3. Pre Model Building Steps: Feature Enginerring and Multicollinearity Check

4. Model Building: Logistic Regression

5. Model Building: Random Forests- Deriving feature importances and building RF Model using best parameters from Grid Search CV

6. Model Building: SVM's- Standardizing Data and then running SVM using best parameters from Grid Search CV

7. Model Building: XGBoost using using best parameters from Random Search CV

8. Model Comparisons/Evaluation using Classification Report and ROC-AUC Curve